Skip to end of metadata
Go to start of metadata

You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 7 Current »

FeaturesPriorityComments

Model management and exposure (MME) service

High

Implement according to procedures/APIs defined by O-RAN alliance 

Reference:  CMCC.AO-2023.06.02-WG2-CR-0019-R1GAP-AIML model management and exposure services-v4.docx

Training ServicesHigh

Implement according to procedures/APIs defined by O-RAN alliance

Reference: INT-2023.05.30-WG2-CR-00050-AIML training use cases-v03.docx

Generic Training Pipeline

HighRequired to support the about services. Create a default generic kubeflow pipeline as part of installation, which the training service can utilize based on the model information provided during training job creation.
AIMLFW optimizations Highinstallation, code refactoring

Automated testing of AIMLFW

HighAutomated scripts to install and test all AIMLFW functions
Advanced Feature selectionMedium
  • Model registration to trigger feature group creation/data request to DME?
  • Support for dynamic change of data source
  • Trigger training only after data is ready in DB
Integrated install with Non-RT RIC/ Near-RT RIC/SMOMedium
Integrate Non-RT RIC and Near-RT RIC AI/ML usecasesMediumNeed to check https://jira.onap.org/browse/DCAEGEN2-3067
Different model deployment optionsMediumCurrently we expose models in the form of zip files that can be deployed. Need to check O-RAN alliance approach.
Model validationLow
Advanced retraining optionsLow
Model Performance monitoringLow

Planned EPICs

  • Generic Training Pipeline
  • New usecases to be supported on AIMLFW
  • Model management and exposure (MME) services
  • Training Services
  • DME Interface enhancements
  • AIMLFW optimizations (installation, code refactoring)
  • Automated testing of AIMLFW
  • No labels